Extend the data set

Question

Hi all,
Please find the below sample scenario where i have to extend the data set.
I am currently working on oracle 10g and while expanding the data set we should exclude the weekends.The sample is only for 2 ids but there can be more ids.
Thanks for all your help.
ID                    ST_DT          END_DT          VAL
====               ========         =======        ========
1                  2/2/2011             2/4/2011       4
1                  2/4/2011             2/8/2011       5
1                  2/8/2011                            6
2                  1/28/2011           2/1/2011        2
2                  2/1/2011                            8
ID                    DT             VAL
====                ========      =======       
1                 2/2/2011           4
1                 2/3/2011           4
1                 2/4/2011           5
1                 2/7/2011           5
1                 2/8/2011           6
1                 2/9/2011           6
2                 1/28/2011          2
2                 1/31/2011          2
2                 2/1/2011           8
2                 2/2/2011           8
2                 2/3/2011           8
2                 2/4/2011           8
2                 2/7/2011           8
2                 2/8/2011           8
2                 2/9/2011           8
select 1 as id,to_date('02/02/2011','MM/DD/YYYY') as st_dt,to_date('02/04/2011','MM/DD/YYYY') as end_dt, 4 as val from dual
union all
select 1 as id,to_date('02/04/2011','MM/DD/YYYY') as st_dt,to_date('02/08/2011','MM/DD/YYYY') as end_dt, 5 as val from dual
union all
select 1 as id,to_date('02/08/2011','MM/DD/YYYY') as st_dt, null as end_dt, 6 as val from dual
union all
select 2 as id,to_date('01/28/2011','MM/DD/YYYY') as st_dt,to_date('02/01/2011','MM/DD/YYYY') as end_dt, 2 as val from dual
union all
select 2 as id,to_date('02/01/2011','MM/DD/YYYY') as st_dt,null as end_dt, 8 as val from dual;

Answer 1

Hi,
I assume that, when end_dt is NULL, you want to use today's date as end_dt, and that the results you posted are the results you want if the query is run on February 10, 2011.
Here's one way:
WITH     got_n_days     AS
     SELECT     id, st_dt, end_dt, val
     ,     NVL ( end_dt
              , TRUNC (SYSDATE)
              ) - st_dt          AS n_days
     FROM     table_x
--     WHERE     ...     -- If you need any filtering, put it here
,     cntr          AS
     SELECT     LEVEL  - 1     AS n
     FROM     (
               SELECT  MAX (n_days)     AS max_n_days
               FROM     got_n_days
     CONNECT BY     LEVEL <= max_n_days - 1
SELECT       d.id
,       d.st_dt + c.n          AS dt
,       d.val
FROM       got_n_days     d
JOIN       cntr          c  ON     c.n     < d.n_days
WHERE       TO_CHAR ( d.st_dt + c.n
              , 'DY'
            , 'NLS_DATE_LANGUAGE=ENGLISH'     -- If necessary
            ) NOT IN ('SAT', 'SUN')
ORDER BY  id
,            dt
;Basically, each row in your original table has to be repeated n times, where n is the number of days between st_dt (included) and end_dt (not included). The first sub-query computes that number, and the main query produces the results by joining your data with a " counter table", cntr, that has one row for every value of n that you might need. That "table" is generated in the second sub-query.
The join produces a row for each day. The WHERE clause eliminates the Saturdays and Sundays from those results.
Thanks for posting the sample data in a useful form! That really helps.
Edited by: Frank Kulash on Feb 10, 2011 2:29 PM